Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra

cognition.com

Post reply on HN
Print view View original

161–170 of 201 posts

Re: Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra

#161

Post by eru »

bayesianbot wrote: post #42
eru wrote: post #27

Earlier quoted context omitted.

> If DeepSeek 4.1 flash is really as good as it's benching, we're probably a month away from 1/3rd of users moving off the closed-weight models in favor of something they have more control over (or is cheaper). Only in the world where the incumbents don't react. Eg if they saw lots of users moving away, they'd drop prices or do something else.

1M cached tokens on deepseek is $0.006, the big labs can't sell anywhere close to this, they have funders expecting returns and huge overhead. btw I've had a ton of fun with the new deepseek today, I was waiting for my OpenAI 5h limit reset and decided to give it some problems for fun, got pretty great results. Tried some harder problems and still got great results. I don't expect it to be Sol class or anything but I…

> 1M cached tokens on deepseek is $0.006, the big labs can't sell anywhere close to this, they have funders expecting returns and huge overhead.

Losing most of your customers tends to sharpen the mind a bit. They could eg stop pushing out the absolute frontier for a while and focus on making what they have run cheaper. Or they go and do more lobbying against China. Or a million other little things that take more than 30 seconds to come up with when writing a HN comment, but less than a week for someone who's smart and paid to do this for a living.

Re: Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra

#162

Post by princevegeta89 »

Saline9515 wrote: post #142
fishtoaster wrote: post #71

Earlier quoted context omitted.

A friend recently pushed me to try out their coding platform, Devin, after I decided to move away from Cursor. I had the same reaction: "What, the con artists from like 2024?" But after some cajoling, I gave it a shot and was pleasantly surprised. I guess they learned their lessons, grew up, and are doing good work now, maybe?

Devin lacks many features and doesn't even have an exec mode. Other than they subsidize the subscription, the added value here is low.

I used Windsurf for quite long and they're basically dead for now after they became Devin.

Everything feels dull and they're always several features behind while Cursor is just killing it every other week.

I'm back to VsCode plus Copilot Pro.

Re: Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra

#163

Post by Bolwin »

FergusArgyll wrote: post #85
airstrafer wrote: post #52

Earlier quoted context omitted.

I guess I don't. Does post-training from another (larger) model not fall under the umbrella of distillation? I'd imagine it leads to the same spiky-ness issues...?

Distilling you don't have the actual model weights of the teacher. All you have are the teachers answers to a lot of questions. You then teach your own smaller model to answer more similarly to the big teacher model. Fine tuning you have the actual model weights of the original model, you then train that model to answer in a different (or better) way.

Distillation requires you to have the actual logits of each token from the teacher model, which in practice means having the model itself.

What you're describing is just synthetic data.

Note Anthropic misused the term in their post about Chinese model distillation, deliberately I assume.

Re: Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra

#165

Post by airstrike »

princevegeta89 wrote: post #162
Saline9515 wrote: post #142

Earlier quoted context omitted.

Devin lacks many features and doesn't even have an exec mode. Other than they subsidize the subscription, the added value here is low.

I used Windsurf for quite long and they're basically dead for now after they became Devin. Everything feels dull and they're always several features behind while Cursor is just killing it every other week. I'm back to VsCode plus Copilot Pro.

Same, but I'm back to just Claude Code and the occasional vscode, though for my latest project I switched to Astra instead

Re: Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra

#166

Post by gpt5 »

mokre wrote: post #126
fallingbananna wrote: post #41

Earlier quoted context omitted.

Those 27.3% are still in the ballpark of modern models: - Sonnet 5 - 12.4% - Luna - 17.3% - Grok 4.6 - 20.3% - Sol - 37.3% - GLM 5.3 - 41.8% - Opus 5 - 51.8%

GLM 5.3 looks strange, because of this Chinese labs benchmaxx moto. So rather they have emergent abilities or... Also a lot of questions to benchmark because opus 5 is completely useless model right now. I think that the main problem with opus that they try to solve context size optimization problem, and that is the main reason why it speaks like alien with only one technical dictionary at hand. So why it is so good?

Terminal Bench 4.0 did not introduce new questions. All tasks were public for a while. If you look at GLM 5.2, which is using the same base model as 5.3, but was released prior to most tasks, it does extremely horribly on terminal Bench 3.0 (4 to 8 times worse than every other model) - source: https://benchlm.ai/benchmarks/terminal-bench-3

Re: Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra

#167

Post by FergusArgyll »

Bolwin wrote: post #163
FergusArgyll wrote: post #85

Earlier quoted context omitted.

Distilling you don't have the actual model weights of the teacher. All you have are the teachers answers to a lot of questions. You then teach your own smaller model to answer more similarly to the big teacher model. Fine tuning you have the actual model weights of the original model, you then train that model to answer in a different (or better) way.

Distillation requires you to have the actual logits of each token from the teacher model, which in practice means having the model itself. What you're describing is just synthetic data. Note Anthropic misused the term in their post about Chinese model distillation, deliberately I assume.

There are multiple kinds of distillation

https://arxiv.org/abs/2106.03310

https://arxiv.org/abs/2207.12106

Re: Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra

#170

Post by fishtoaster »

klardotsh wrote: post #138
fishtoaster wrote: post #71

Earlier quoted context omitted.

A friend recently pushed me to try out their coding platform, Devin, after I decided to move away from Cursor. I had the same reaction: "What, the con artists from like 2024?" But after some cajoling, I gave it a shot and was pleasantly surprised. I guess they learned their lessons, grew up, and are doing good work now, maybe?

Devin CLI is easily the worst-in-class coding harness I've ever been subjected to using, rife with bugs (up to and including dropping answers to the question tool), so I can't say I'm exactly inspired to try anything else the company produces.

Can't say I've tried the CLI. I've mostly focused on the cloud agents, which I was explicitly looking for. I compared against cursor's cloud agents, ampcode, and hoplite, and came out surprisingly enjoying devin.

I will say that the lack of parity between Devin cloud and Devin desktop is downright embarrassing. It's very clear that the latter is a thinly-reskinned Windsurf. A visually similar UI with vastly different capabilities. Definitely a black mark on the whole thing.

Post reply on HN